It’s been a hit since I read up on it, but from my understanding alpha fold 2 tended to be more accurate, while roseTTA was able to predict complexes. However I know that recently alpha fold 2 has a multimer colab notebook, not sure how good it is tho. Both are pretty good tools and probably work well when you take into accounts from both
It’s been a hit since I read up on it, but from my understanding alpha fold 2 tended to be more accurate, while roseTTA was able to predict complexes. However I know that recently alpha fold 2 has a multimer colab notebook, not sure how good it is tho. Both are pretty good tools and probably work well when you take into accounts from bothMore
I'm on the software & infrastructure side of things, so perhaps don't have the depth of the answer you're looking for, and would be happy to be corrected. But as far as I understand from the scientists on our team is that "it depends". Both can get results that are remarkably close to one another, but both can also disagree, so it's hard to say what's "better" - it depends on what you consider the ground truth for your experiments, and a number of other factors such as the length of the sequences you're working with. Both are able to predict structure from sequence, recover sequence from structure, as well as "hallucinate" structure and sequence, so it can be helpful to use both to validate your results and increase your confidence. I assume you've seen the ColabFold notebooks , so it should be easy to compare them against each other for your use case.
I'm on the software & infrastructure side of things, so perhaps don't have the depth of the answer you're looking for, and would be happy to be corrected. But as far as I understand from the scientists on our team is that "it depends". Both can get results that are remarkably close to one another, but both can also disagree, so it's hard to say what's "better" - it depends on what you consider the ground truth for your experiments, and a number of other factors such as the length of the sequences you're working with. Both are able to predict structure from sequence, recover sequence from structure, as well as "hallucinate" structure and sequence, so it can be helpful to use both to validate your results and increase your confidence. I assume you've seen the ColabFold notebooks , so it should be easy to compare them against each other for your use case.More
Thanks so much for this! I have actually not seen this. I’m not really in the bioinformatics field, but I’ve always had an interest in biology as a physics and chemistry double major with some bio courses as well, so figured it would be a good hobby to pick up on the side.More
VOTE
VOTE
VOTE
VOTE